1. 03-12-12 04:21 PM #1
    /\ox
    /\ox's Avatar Become A Pro!
    Join Date: 03-05-12
    Posts: 47
    Betpoints: 185

    finding balance between played games and other data

    Hello,

    I'm messing with NBA model, so I wonder what's more "important"/I should give more weight to - results and performance in the last 5 games (i.e current form) OR home/away, rest days, last 8 matches between teams, injuries and power ranking.

    Tonight's Hornets vs Bobcats matchup Bobcats took lead gamewise over Hornets (because Bobcats' big win over Magic) but loses its momentum in light of other data (2-6 vs Hornets; Hornets are home, away record is 2-19 etc). For example, my formula's first part (last 5 games) gives 9,2 points to Bobcats and 6,3 points to Hornets, how I should treat other data now, compared to this first calculation.

    Any directions would be very helpful, thank you!
    Last edited by /\ox; 03-13-12 at 06:49 AM.
    Reply With Quote Reply With Quote
  2. 03-13-12 04:51 AM #2
    /\ox
    /\ox's Avatar Become A Pro!
    Join Date: 03-05-12
    Posts: 47
    Betpoints: 185

    BUMP - my question sucks really hard?
    Reply With Quote Reply With Quote
  3. 03-13-12 06:40 AM #3
    Dark Horse
    Deus Ex Machina
    Dark Horse's Avatar Become A Pro!
    Join Date: 12-14-05
    Posts: 13,764
    Betpoints: 2

    No, it doesn't. The answer is within your reach. It just requires work. Or did your really expect others to do that work for you? The question is the 1% inspiration. The 99% perspiration comes next.
    Reply With Quote Reply With Quote
  4. 03-13-12 06:51 AM #4
    /\ox
    /\ox's Avatar Become A Pro!
    Join Date: 03-05-12
    Posts: 47
    Betpoints: 185

    Dark Horse,

    no I didn't expect someone to do my work. I'm working on it already and trying to figure out the right answers
    Just asked for some directions or general guidelines and/or user's experiences.

    ty!

    /\ox
    Reply With Quote Reply With Quote
  5. 03-13-12 09:12 AM #5
    mathdotcom
    mathdotcom's Avatar Become A Pro!
    Join Date: 03-24-08
    Posts: 11,689
    Betpoints: 1943

    Quote Originally Posted by Dark Horse View Post
    No, it doesn't. The answer is within your reach. It just requires work. Or did your really expect others to do that work for you? The question is the 1% inspiration. The 99% perspiration comes next.
    ??? No it does not just require work

    In general short term trends do not have any statistical significance, but you can try backtesting.
    Reply With Quote Reply With Quote
  6. 03-14-12 12:22 AM #6
    Dark Horse
    Deus Ex Machina
    Dark Horse's Avatar Become A Pro!
    Join Date: 12-14-05
    Posts: 13,764
    Betpoints: 2

    Shorter term data have their place in weighted averages.
    Reply With Quote Reply With Quote
  7. 03-14-12 02:19 PM #7
    Justin7
    Justin7's Avatar Become A Pro!
    Join Date: 07-31-06
    Posts: 8,577
    Betpoints: 1506

    Try different approaches, and test to see which one is more predictive.

    The recent data will typically work better if there was an injury. If there was no injury, season-long works better. The best approach is to use season-long data, and find ways to adjust for injuries/roster changes.
    Reply With Quote Reply With Quote
  8. 03-15-12 02:16 AM #8
    /\ox
    /\ox's Avatar Become A Pro!
    Join Date: 03-05-12
    Posts: 47
    Betpoints: 185

    backtesting right now, thank you guys!
    Reply With Quote Reply With Quote
Top
« Previous Thread | Handicapper Think Tank | Next Thread »